A Robust Method for Detecting Copy-Move Image Forgery Using Stationary Wavelet Transform and Scale Invariant Feature Transform

Copy-move image forgery is one type of image forgery where a part of the image is copied and then it is pasted in the same image to hide or add some important object(s) within the image. Most image forgery detection models are unable to detect forgery in the image if the copied portion has noise or it is rotated or scaled before pasting. The purpose of this paper is to propose a robust and efficient detection technique for this kind of image forgery. Firstly, the image is converted to grayscale. Then two-level Stationary Wavelet Transform (SWT) is used to decompose the grayscale image into four parts and Scale Invariant Feature Transform (SIFT) algorithm is used to extract the key-points from the approximate part of the decomposed image. Later, the matched pairs of key-points are identified. Then matched pairs of key-points are clustered using several linkage methods. Clustered key-points are compared to take the decision whether the image is tampered or not. In the post-processing step, false positive matches are removed using Random Sample Consensus (RANSAC). The proposed model shows 93% accuracy over a certain dataset of images.